Research Report
DOES DIVERSITY HURT
DEMOCRACY?
M. Steven Fish and Robin S. Brooks
M. Steven Fish is associate professor of political science at the University of California, Berkeley. He is a coauthor of Postcommunism and
the Theory of Democracy (2001). Robin S. Brooks is a Ph.D. candidate
at the University of California, Berkeley. She is writing her dissertation on ethnic identity and political change in southeastern Europe.
While reporting recently on his organization’s annual survey of political rights and civil liberties around the world, Freedom House
president Adrian Karatnycky repeated the claim—heard often in academic and public discourse alike—that ethnic diversity hinders open
politics. Reviewing the findings of the 2001 survey, he concluded that
“democracy has been significantly more successful in monoethnic societies than in ethnically divided and multiethnic societies.”1
In saying this, Karatnycky was hardly being provocative or
counterintuitive. A number of eminent political scientists have seen
diverse societies as disadvantaged when it comes to democratization.2
According to many observers, ethnic differences divide society and
make compromise and consensus difficult. Heterogeneity poses the risk
of intercommunal violence, which can quickly undermine open politics. What is more, political parties and other organizations coalesce
more readily around ethnic than other identities. Political entrepreneurs therefore have an incentive to play on such divisions and to
neglect efforts to mobilize citizens around civil rights and class concerns. 3 And in a particularly ironic twist, well-meant efforts to defuse
ethnic conflict can take the form of elite bargains, made amid political
openings, that later block further democratization.4
Empirical evidence seems abundant. Writing in the wake of the Soviet demise, Donald L. Horowitz observed: “Democracy has progressed
furthest in those East European countries that have the fewest serious
ethnic cleavages (Hungary, the Czech Republic, and Poland) and progressed more slowly or not at all in those that are deeply divided
Journal of Democracy Volume 15, Number 1 January 2004
M. Steven Fish and Robin S. Brooks
155
(Slovakia, Bulgaria, Romania, and of course the former Yugoslavia).”5
Many other examples are available. In Asia, the relative smoothness of
democratization in monoethnic South Korea contrasts with the rocky
course of regime change in collaged Indonesia. In Latin America, democracy seems more robust in Chile and Costa Rica, which are relatively
homogeneous, than in fractionalized Peru and Guatemala.
The notion that greater ethnic homogeneity facilitates democracy
simply seems to make sense. Who doubts that maintaining popular
rule is going to be easier in Poland and Slovenia than it is in Macedonia
and Bulgaria? Is it any wonder that ethnically uniform Greece has had
a more successful experience with open politics than has Turkey, a
fifth of whose population is self-consciously and assertively Kurdish?
And does anyone really expect democracy to take firm root in Bosnia,
however long the UN stages elections and stands between hostile
groups?
Yet closer inspection reveals surprisingly scanty evidence that diversity countervails open politics. Here we present the findings of
elementary cross-national analysis.
Much academic writing on democracy’s determinants assumes or avers
that social heterogeneity dims democracy’s prospects. But the effects of
social composition often go untested. The main reason is very likely
that fractionalization—the degree to which a society is divided up into
various distinct groups—is hard to measure. Group identities are complex and contested; quantifying them is problematic. Social science
does not yet, and perhaps never will, have uncontroversial measures of
social fractionalization.
Until recently, scholars who have sought to assess the effects of social diversity in cross-national analyses have had only a handful of
sources from which to draw. Some are badly dated and lack data on
certain countries that have emerged from colonial or communist rule in
the last few decades. What is more, studies of fractionalization typically treat linguistic distinctions alone as the basis for assessing group
membership.6 While this data has yielded some useable measures, its
general inadequacy is well known among scholars.7
In addition to scoring for “fractionalization,” some experts gauge
the diversity of a society by the relative size within it of its largest
group: The higher the percentage of the total national population accounted for by this group, the closer the society is to homogeneity. In
its annual publications, Freedom House provides information on the
size of what it defines as ethnic groups. Freedom House’s numbers on
ethnic composition are vulnerable to criticism. Perhaps the most serious deficiency is the absence of a uniform criterion to define ethnicity.
For some countries, such as South Africa, ethnicity is assessed in terms
of race. The country’s ethnic profile is characterized as “75 percent
black” and “14 percent white.” The rest, presumably, are mostly “East
156
Journal of Democracy
Indians” and mixed-heritage “Cape Coloureds.” In other cases, subracial
distinctions of tribe and language figure in the Freedom House account. In material on Namibia, for example, the largest group is not
“black” but Ovambo—a group that forms about half the population
and is distinguished from other and smaller indigenously African
groups such as the Kavango. Were the same types of distinctions applied in the case of South Africa, that country’s diversity would be
more obvious in the numbers. In some cases, religious identity is included in the assessment. Thus the Philippines is judged to be “91.5
percent Christian Malay” and “4 percent Muslim Malay.” If language
were treated as the main criterion for evaluation, the Philippines would
be considered highly diverse.
Thus for Freedom House, “ethnicity” may refer to race (meaning
physiognomy and skin color), language, religion, or some combination
thereof. Any attempt to assess ethnic composition must take on the
thorny task of saying just which criteria are supposed to count, how
they can be measured, and why they matter. 8 Assessments that focus on
language, as ethnic-fractionalization indices typically do, have the advantage of hewing to one more or less clear criterion, but therein lies
one of their drawbacks as well: They miss much of what constitutes
sociocultural difference. The numbers that Freedom House uses for
“ethnicity,” while sometimes transcending the limits of an exclusively
linguistic focus, do not stick with a single criterion, which limits their
usefulness when it comes to cross-national analysis.
Fortunately, there has lately been a breakthrough in the effort to
assess social composition. Alberto Alesina and his colleagues offer scores
on “ethnic” (by which they mean, for the most part, racial), linguistic,
and religious fractionalization for nearly all the world’s countries.9 This
source (hereafter referred to as “Alesina data”) provides a useful instrument for measuring social homogeneity and heterogeneity. Since the
main components of what is often called “ethnicity” are disaggregated,
the influence of each component may be assessed separately. The data
make possible a differentiated assessment of the influence of social
diversity on political regime. The scores range from zero to one, with
lower scores representing lower fractionalization.
We also use the data on ethnicity that Freedom House publishes in
its own annual reports. Whatever its shortcomings, the Freedom House
material furnishes another way to assess social composition, and also
forms the basis for Karatnycky’s claim. We treat the size of the largest
ethnic group as the measure of “ethnic homogeneity.” Karatnycky dubs
countries in which the largest group accounts for two-thirds or more of
the population “monoethnic”; all other countries he terms
“multiethnic.” 10 He holds that greater uniformity, understood as stronger numerical predominance by the largest ethnic group, creates better
conditions for democracy. We test this hypothesis using the numbers
M. Steven Fish and Robin S. Brooks
157
on ethnic composition that Freedom House published in a recent annual report. 11
Analyzing the Effects of Diversity
We seek to evaluate the effect of social homogeneity or heterogeneity on political regime by examining countries with at least a quarter of
a million inhabitants as of the year 2000. To assess political regime, we
use Freedom House’s freedom ratings (hereafter FH ratings), which are
published annually for each country in the world. We regard the ratings
as a valuable—arguably the best—source of cross-national data on political regimes. Each country’s FH rating is an average of the scores that
the Freedom House staff assigns it in the areas of “political rights” and
“civil liberties,” respectively. The scale ranges from 1 (most free) to 7
(least free). To make our presentation more intuitive, we reverse the
scale so that a higher number means a higher degree of openness (in
other words, 7 represents greatest freedom, 1 least freedom). We score
countries using an average of their ratings over the five most recent
annual surveys (those issued between 1998 and 2002).12 These scores
serve as measures for “political regime”—the dependent variable that
we are trying to explain.
Assessing the effects of social fractionalization requires controlling
for several other factors that are also widely regarded as determinants of
political regime. The first and most obviously indispensable is level of
economic development, which we measure as Gross Domestic Product
per capita (GDPpc).13 Scholars have long held that higher economic
development is associated with less social conflict, higher political
sophistication, and broader social support for popular rule.14
Three other controls—the predominance of Islam as the main religion,
a British colonial heritage, and membership in the Organization of Petroleum Exporting Countries (OPEC)—are coded as dummy variables (that
is, simply present or not). Some scholars have noted what appears to be a
deficit of democracy in the Muslim world.15 According to their analysis,
the fusion of temporal and spiritual authority in Islamic thought, the
subordination of women, and a culture of intolerance predispose Muslim
societies to authoritarianism. Conversely, argue other experts, a British
colonial heritage is most often good for democracy.16 The British, in this
view, handed down traditions of law, parliamentarism, and civil-service
professionalism that left their former colonies in a better position to sustain open rule than the former colonies of other European powers.
The presence of abundant raw materials—especially oil—has sometimes been regarded as a harbinger of woe for democratic prospects. Oil
wealth may buy public quiescence, finance a large security apparatus to
repress opposition, and promote a type of economic growth that does
not engender genuine modernization.17
.36
N
.54
4
.06
-1.80***
(0.47)
5.06***
(0.27)
5
.36
161
2.34***
(0.27)
-0.07
(0.45)
-4.04***
(1.06)
6
.55
-1.42**
(0.50)
0.13
(0.29)
-1.50***
(0.28)
2.03***
(0.27)
0.05
(0.41)
-2.46*
(1.10)
7
.00
0.76
(0.61)
.36
166
2.38***
(0.23)
0.81
(0.50)
-4.54***
(0.83)
8
.54
-1.47**
(0.52)
0.25
(0.31)
-1.52***
(0.29)
2.07***
(0.24)
-0.48
(0.47)
-2.40*
(0.98)
9
.03
0.015*
(0.006)
3.10***
(0.42)
10
11
.34
138
2.44***
(0.31)
-0.0002
(0.006)
-4.52***
(0.95)
HYPOTHESIZED DETERMINANTS
3.98***
(0.31)
ON
12
.53
-1.61**
(0.51)
0.16
(0.33)
-1.28***
(0.29)
2.34***
(0.28)
-0.006
(0.005)
-3.22**
(1.06)
Note: OLS regressions. Entries are unstandardized regression coefficients with White-corrected robust standard errors in parentheses. *p<0.05; **p<0.01; ***p<0.001
Sources: For ethnic, linguistic, and religious fractionalization: Alberto Alesina, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat, and Romain Wacziarg, “Fractionalization,” National
Bureau of Economic Research Working Paper No. 9411, January 2003; for ethnic homogeneity, Freedom House, Freedom in the World: The Annual Survey of Political Rights and Civil Liberties
1999–2000 (Lanham, Md.: Rowman & Littlefield, 2000); for economic development, CIA World Factbook 2000 (Washington, D.C.: Brassey’s, 2000).
166
-1.46**
(0.52)
OPEC
Member
Adj. R 2
0.16
(0.29)
2.13***
(0.28)
0.11
(0.50)
Former
British Colony
2.22***
(0.30)
-0.62
(0.57)
-1.46***
(0.28)
.11
-2.63***
(0.53)
3
FREEDOM HOUSE RATINGS
-2.86*
(1.22)
OF
Islam
Economic
Development
Ethnic
Homogeneity
Religious
Fractionalization
Ethnic/Racial
Fractionalization
Linguistic
Fractionalization
Constant
2
-3.33**
(1.24)
1
5.52***
(0.31)
TABLE 1—R EGRESSIONS
158
Journal of Democracy
M. Steven Fish and Robin S. Brooks
159
A disclaimer is in order: We are well aware that these controls do not
exhaust the possible determinants of political regime. We intend merely
to test the impact of social fractionalization on political regime under a
set of rudimentary controls, not to provide a comprehensive evaluation
of the causes of cross-national variation in political regime.
That said, the 12 columns of Table 1 present the results of our regression analyses. The models in columns 1 through 9 show the analyses
using the Alesina fractionalization data in three different dimensions:
ethnic (1 through 3); linguistic (4 through 6); and religious (7 through
9). The Alesina data set lacks scores on ethnicity for Maldives and
Yemen, on language for Cuba, El Salvador, Haiti, Maldives, Rwanda,
and Yugoslavia, and on religion for Maldives and Yugoslavia. The analyses of the effects of ethnic, linguistic, and religious fractionalization
therefore include 166, 161, and 166 observations, respectively. Models
10 through 12 present the results of analyses that use the size of the
largest ethnic group as a proportion of the total population as the indicator for ethnic homogeneity. Our source for relative ethnic-group sizes
is the published Freedom House data, which covers 138 of the world’s
168 countries with populations of at least a quarter-million each.
The analyses strongly support the hypothesis that economic development affects political regime. Greater wealth is associated with more
open government. So too do the regressions show that predominantly
Muslim countries are marked underachievers in democracy. Resource
abundance also seems to frustrate open politics. A British colonial heritage, by contrast, does not affect democracy’s prospects either way.
Our main concern, however, is the effect of social diversity on democracy. Model 1, which presents a simple bivariate regression, shows
that ethnic fractionalization is negatively correlated with political openness. The correlation is not high, however, and the ethnicity variable is
not robust to the inclusion of controls, as is evident in models 2 and 3.
Model 2 adds the control for economic development and model 3 includes the other controls as well. In neither of these models is ethnic
fractionalization statistically or substantively significant.
Analysis of the effect of linguistic fractionalization, shown in models 4 through 6, yields similar results. The correlation is modest in the
bivariate regression, and in neither specification that includes controls
is linguistic fractionalization statistically significant. In the regressions
that test the effects of religious fractionalization, shown in models 7
through 9, diversity again fails to hold up as a predictor of political
regime. In the bivariate regression, shown in model 7, and in the multivariate regression that adds the control for economic development alone
(shown in model 8) the coefficient for fractionalization is actually positive. Greater religious fractionalization is therefore associated with better
FH scores, though the relationship is not statistically significant.
Likewise, applying the data on ethnic homogeneity yields no evi-
160
Journal of Democracy
dence of a sturdy, sizeable link between social homogeneity and democracy. Contrary to Karatnycky’s claim, we see that the relative size
of the largest ethnic group is not a good predictor of democratic achievement. In the bivariate regression (model 10) the correlation between the
size of the largest ethnic group and the FH ratings is paltry. Models 11
and 12 show that when controls are added the link between homogeneity and democracy actually becomes negative, though the relationship
is not statistically significant. Greater ethnic homogeneity is not associated with more open political regimes.
Some scholars have suggested that social fractionalization may be important, but with complicated and nonlinear or curvilinear effects that
might not appear in regression analyses. Paul Collier and Anke Hoeffler
argue that countries in which one finds “ethnic dominance”—meaning
that a single ethnic group makes up 45 to 90 percent of the population—
are more prone to major civil conflict than either those highly
homogeneous countries where one group forms more than 90 percent of
the population or those highly fractionalized countries where each group
comprises less than 45 percent of the population.18 Collier and Hoeffler
focus on ethnic composition’s effect on conflict rather than on democracy,
but the logic of the claim may be applied readily to democracy as well.
If a hazardous intermediate zone between very high homogeneity and
very high fractionalization actually exists, one should see evidence of a
U- or J-shaped curve in diagrams that plot the correlation between FH
scores and four indicators of social diversity. We assembled scatter plots
using these four indicators.19 These plots reveal no such telltale U- or Jshaped arc. They do, however, illustrate what is evident in the regressions:
There is scant correlation between social diversity and political regime.
In sum, the degree of diversity is not shown to influence democracy’s
prospects. What social scientists call the “null hypothesis”—the proposition that social diversity has no appreciable effect on political
regime—is one that in this case we cannot readily reject.
Diversity in Nonwealthy Democracies
We can extend the investigation and check our results by narrowing
the universe and briefly examining a set of cases that share certain
traits. It may be especially useful to focus on the developing world.
There, multiethnicity is often viewed as presenting especially daunting challenges. Perhaps Belgium, Switzerland, and Canada, with their
wealth and their traditions of popular rule, can maintain democratic
regimes despite diversity, while in developing countries this might be
harder.
To separate wealthy from developing countries we use a threshold
that has recently become a benchmark in writings on democracy’s requisites. In an influential article, Adam Przeworski and Fernando Limongi
.46
.46
.79
.74
.41
.40
.42
.43
.20
.51
.62
.42
.42
.59
.32
.69
.37
.63
.27
.24
.31
IN
.44
.40
LINGUISTIC
FRACTIONALIZATION*
.79
.22
.41
.30
NA
.04
NA
.49
.07
.81
.11
.58
.32
.84
.37
.70
.35
.84
.17
.42
.43
RELIGIOUS
FRACTIONALIZATION*
.55
.21
.60
.60
.08
.31
.36
.50
.79
.33
.62
.56
.41
.18
.08
.66
.55
.31
.24
A
72
71
LARGEST ETHNIC GROUP AS
PERCENTAGE OF THE
TOTAL POPULATION**
NA
30
95
85
71
73
94
64
49
72
76
57
80
50
90
50
NA
92
89
L OW-INCOME D EMOCRACIES (< US$6,000 Annual GDPpc)
*0=lowest, 1=highest fractionalization; **100=greatest homogeneity
Sources: For ethnic, linguistic, and religious fractionalization: Alberto Alesina et al., “Fractionalization”; for largest group as a percentage of population, Freedom in the World: The Annual Survey of Political
Rights and Civil Liberties, 1999–2000.
COUNTRIES
Benin
Bolivia
Botswana
Bulgaria
Cape Verde
Dominican Republic
El Salvador
Estonia
Guyana
India
Jamaica
Latvia
Lithuania
Mali
Mongolia
Namibia
Papua New Guinea
Philippines
Romania
Mean: Low-Income
Democracies
Mean: World
ETHNIC
FRACTIONALIZATION*
TABLE 2—FRACTIONALIZATION
M. Steven Fish and Robin S. Brooks
161
162
Journal of Democracy
reported that above an income of $6,000 per capita no major country
that has become a democracy has ever reverted to authoritarianism. 20
The number provides a reasonable threshold for distinguishing between wealthy and developing countries. As of 1998, income per capita
was less than $6,000 per year in 1998 U.S. dollars at Purchasing Power
Parity in 107 of the 168 countries with populations of more than a
quarter-million. (These 107 were home to about three-quarters of the
people in the world.) Within this realm of developing countries, 19
received FH ratings in each of the five annual surveys between 1998
and 2002 that placed them in Freedom House’s category of “Free”
polities. They are exceptions to the generalization that open politics
is a luxury that only the rich can afford. These low-income, relatively
liberal democracies, which are listed in Table 2 on the preceding page,
make up a variegated club. There is no hint of geographical concentration. While Protestant and Catholic countries are especially well
represented, all other major religious traditions have a place in this
category as well. Orthodox Christianity predominates in Bulgaria and
Romania, Hinduism in India, Buddhism in Mongolia, Islam in Mali,
and traditional animism (Vodou) in Benin. The colonial heritage of
the countries in this group is also obviously anything but uniform.
Of even greater interest than the heterogeneity of the set of countries
as a whole is the diversity on display within each of them. Table 2
presents the data on social fractionalization and ethnic homogeneity
for each country. The overall picture reinforces the findings of the statistical analysis. The average ethnic fractionalization score for the
developing-world states that Freedom House rates as “Free” is identical
to the global average. Linguistic fractionalization is moderately higher
than the global mean. Religious fractionalization is virtually identical
to the average for the world as a whole. And finally, the relative size of
the largest ethnic group is nearly the same when averaged across the
nonwealthy democracies as it is when averaged across the world as a
whole. Open politics is not tethered to social uniformity.
Why Doesn’t Diversity Matter?
An important question arises from the findings: Why does social
fractionalization not matter for democracy? In order to address the matter, one must ask: Why should diversity hinder democracy? One argument
associates greater diversity with a higher propensity for major civil
conflict. Some scholars even use fractionalization as a proxy for the
degree of conflict in society, operating on the assumption that higher
fractionalization automatically translates into more conflict.21 Since
there is obviously good reason to regard violence as an antagonist of
open rule, if higher fractionalization does mean more violence, one
would indeed expect diversity to make things hard for democracy.
M. Steven Fish and Robin S. Brooks
163
Several recent studies, however, challenge the soundness of the idea
that heterogeneity is linked to violence. James Fearon and David Laitin
report, “It appears not to be true that a greater degree of ethnic or religious diversity—or indeed any particular cultural demography—by
itself makes a country more prone to civil war.” Fearon and Laitin note
that their finding “runs contrary to a common view among journalists,
policy makers, and academics, which holds ‘plural’ societies to be especially conflict-prone due to ethnic or religious tensions or
antagonisms.”22
In a major review of the politics of identity in Africa, Crawford Young
argues that the recent growth of violence there does not spring primarily from social heterogeneity, even if the latter often becomes an axis
of cleavage once violence breaks out. Young notes that “once armed
conflict is interwoven with politics, identity is virtually certain to become part of the larger patterns of confrontation.” Still, the new patterns
and intensity of conflict “have nothing to do with religion, ethnicity,
and race.” In fact, “cultural pluralism alone is not the prime determinant
[of conflict]; countries that have escaped disorder are no less diverse
than those in which armed conflict has erupted.”23
While much more research needs to be done on the topic of fractionalization and conflict (now a vigorously debated one among political
scientists and economists), the present state of understanding raises
the possibility that there is little or no connection between the two
phenomena.
Another basis for regarding fractionalization as an antagonist of open
politics is diversity’s supposedly pernicious effect on economic performance. In a celebrated article, William Easterly and Ross Levine reported
that high ethnic fractionalization provided a powerful explanation for
slow growth in Africa. 24 If diversity undermines growth and if growth
promotes democratization, one might expect heterogeneity to countervail political opening.
Easterly and Levine’s 1997 article furnishes sound evidence, but
the question of whether diversity spells adversity remains open, for
much remains unknown and scholars as yet have no firm grasp of
whether and how social heterogeneity affects politics. In subsequent
work Easterly has qualified his earlier conclusions, arguing that “highquality institutions, such as rule of law, bureaucratic quality, freedom
from government expropriation, and freedom from government repudiation of contracts mitigate the adverse economic effects of ethnic
fractionalization.”25 Other writers, moreover, have found no link between high diversity and low growth. 26 While we have not undertaken
systematic study of the problem, in preliminary analysis we too have
found little evidence for a strong link between social diversity and
economic performance. 27
In sum, even though our knowledge is far from complete, we do know
Journal of Democracy
164
enough to hazard that the common assumptions about heterogeneity
promoting conflict and stifling growth should be judged “not proven.”
Implications for Politics
If the empirical evidence plainly pointed to the conclusion that
multiethnicity dims democracy’s prospects, the finding might be a “hard
truth” for democratic idealists but would nevertheless merit sober acceptance. But this claim is not demonstrably true, and indeed is not
even well supported by straightforward cross-national analysis.
This is not to say that our rudimentary examination provides the last
word on the matter. On the contrary, it is intended merely as a single and
preliminary contribution to the broader effort, now gaining momentum
among social scientists, to assess the political influence of fractionalization. Our findings, which are based on analysis of some high-quality,
highly differentiated new data, provide grounds for doubt about the
idea that monoethnic societies have an edge when it comes to founding
and preserving democratic rule.
This is no mere intellectual debate among experts. While the idea that
monoethnic societies have advantages may be the brainchild of honest
scholars and commentators, its logically implied converse—that ethnically fragmented societies are intrinsically troubled—is ruthlessly
manipulated by undemocratic rulers the world over. Few excuses for
authoritarianism are trotted out more frequently than the claim that multiform societies need a strong hand to prevent all hell from breaking loose.
Singapore’s longtime strongman, Lee Kwan Yew, has argued for decades
that his country’s diversity makes democracy a bad fit. Highhanded premier Mohamad Mahathir of Malaysia has long made the same argument
about his society. The president of Uganda, Yoweri Museveni, continues
to ban multipartism on the grounds that allowing it will lead to interethnic
war. China’s rulers, having long since abandoned state ownership and rule
on behalf of the poor as rationalizations for authoritarianism, now justify
their dictatorship in part by invoking China’s multinational character and
potential for social strife. Dictators in Burma as well as all five of the postSoviet Central Asian republics make precisely the same argument.
According to their logic, democracy is inappropriate in diverse societies precisely because it is unsustainable; political opening will only spark
mass conflict and thereby undermine even the scant rights and security
that the populace enjoys under authoritarianism. Western leaders mostly
accept this line, at least tacitly, and expect less from multiethnic polities.
Examining the evidence reveals the tenuousness of such reasoning. If a
robust connection between social homogeneity and political openness does
not exist in global perspective, and if a substantial number of the developing world’s relatively liberal democracies are decidedly multiethnic, then
the number of plausible pretexts for despotism falls by one.
M. Steven Fish and Robin S. Brooks
165
NOTES
1. Adrian Karatnycky, “The 2001 Freedom House Survey: Muslim Countries
and the Democracy Gap,” Journal of Democracy 13 (January 2002): 107.
2. Arend Lijphart, Democracy in Plural Societies (New Haven: Yale University
Press, 1977); Robert A. Dahl, Polyarchy: Participation and Opposition (New Haven: Yale University Press, 1971); Alvin Rabushka and Kenneth A. Shepsle, Politics
in Plural Societies (Columbus, Ohio: Merrill, 1972); David Welsh, “Domestic
Politics and Ethnic Conflict,” in Michael E. Brown, ed., Ethnic Conflict and
International Security (Princeton: Princeton University Press, 1993), 43–60.
3. Donald L. Horowitz, Ethnic Groups in Conflict (Berkeley: University of
California Press, 1985).
4. Courtney Jung and Ian Shapiro, “South Africa’s Negotiated Transition,”
Politics and Society 23 (September 1995): 269–308.
5. Donald L. Horowitz, “Democracy in Divided Societies,” Journal of Democracy 4 (October 1993): 18–38.
6. Charles L. Taylor and Michael C. Hudson, World Handbook of Political and
Social Indicators, 2 nd ed. (New Haven: Yale University Press, 1972); Erik V.
Gunnemark, Countries, Peoples, and Their Languages: The Geolinguistic Handbook (Dallas: SIL International, 1991); Barbara F. Grimes, ed., Ethnologue
Languages of the World, 14 th ed. (Dallas: SIL International, 2000).
7. See, for example, David D. Laitin and Daniel N. Posner, “The Implications of
Constructivism for Constructing Ethnic Fractionalization Indices,” Newsletter of
the Comparative Politics Section of the American Political Science Association 12
(Winter 2001): 13–17.
8. See Ashutosh Varshney, “Nationalism, Ethnic Conflict, and Rationality,”
Perspectives on Politics 1 (March 2003): 85–99.
9. Alberto Alesina, Arnaud Devleeschauwer, William Easterly, Sergio Kurlat,
and Romain Wacziarg, “Fractionalization,” National Bureau of Economic Research
Working Paper No. 9411, January 2003. This study is not the only potentially
fruitful recent effort to overcome the limitations of the older measures of social
diversity. See also, for example, Daniel N. Posner, “Ethnic Fractionalization in
Africa: How Should It Be Measured? What Does It Explain About Economic
Growth?” Paper presented at the 1999 Annual Meeting of the American Political
Science Association, Atlanta, Ga., September 1999.
10. Adrian Karatnycky, “The 2001 Freedom House Survey,” endnote 1.
11. Freedom in the World: The Annual Survey of Political Rights and Civil
Liberties 1999–2000 (New York: Freedom House, 2000).
12. See Freedom House’s “Annual Survey of Freedom, Country Ratings,” at
www.freedomhouse.org.
13. CIA World Factbook 2000 (Washington, D.C.: Brassey’s, 2000). In the
regression analysis we use log GDP per capita in 1998 at Purchasing Power Parity
as the measure of economic development.
14. Seymour Martin Lipset, Political Man (Garden City, N.Y.: Doubleday,
1960); Valerie Bunce, “Comparative Democratization: Big and Bounded Generalizations,” Comparative Political Studies 33 (August–September 2000): 703–34.
166
Journal of Democracy
15. Adrian Karatnycky, “The 2001 Freedom House Survey,” endnote 1; Jan
Goodwin, Price of Honor: Muslim Women Lift the Veil of Silence in the Islamic
World (New York: Penguin, 1995); Ann Elizabeth Mayer, Islam and Human Rights:
Tradition and Politics (Boulder, Colo.: Westview, 1998); M. Steven Fish, “Islam
and Authoritarianism,” World Politics 55 (October 2002): 4–37.
16. Myron Weiner, “Empirical Democratic Theory,” in Myron Weiner and
Ergun Özbudun, eds., Competitive Elections in Developing Countries (Durham,
N.C.: Duke University Press, 1987): 3–37; Seymour Martin Lipset, “The Centrality
of Culture,” in Larry Diamond and Marc F. Plattner, eds., The Global Resurgence
of Democracy, 2 nd ed. (Baltimore: Johns Hopkins University Press, 1996), 150–53.
17. Michael Ross, “Does Oil Hinder Democracy?” World Politics 53 (April
2001): 325–61.
18. Paul Collier and Anke Hoeffler, “Greed and Grievance in Civil War,” October 2001. At http://econ.worldbank.org/docs/1109.pdf. Site consulted April 2003.
A similar argument, based on formal logic, is made in Thomas P. Tangeras and NilPetter Lagerlof, “Ethnic Diversity and Civil War,” Research Institute of Industrial
Economics, Stockholm, Working Paper No. 589 (2003). www.iui.se/wp/Wp589/
IUIWp589.pdf.
19. These plots could not be reproduced at sufficient size within the pages of
the Journal of Democracy, and so have been posted instead on the Journal’s
website at www.journalofdemocracy.org/articles/Fish-15-1.pdf.
20. Adam Przeworski et al., “What Makes Democracies Endure?” Journal of
Democracy 7 (January 1996): 39–55.
21. Anthony Annett, “Social Fractionalization, Political Instability, and the Size
of Government,” IMF Staff Papers 48, 3 (2001). www.imf.org/External/Pubs/FT/
staffp/2001/03/pdf/annett.pdf.
22. James D. Fearon and David D. Laitin, “Ethnicity, Insurgency, and Civil
War,” American Political Science Review 97 (February 2003): 75. See also Barbara
Harff, “No Lessons Learned from the Holocaust? Assessing Risks of Genocide and
Political Mass Murder Since 1955,” American Political Science Review 97 (February 2003): 57–73.
23. Crawford Young, “Deciphering Disorder in Africa: Is Identity the Key?”
World Politics 54 (July 2002): 540.
24. William Easterly and Ross Levine, “Africa’s Growth Tragedy: Policies and
Ethnic Divisions,” Quarterly Journal of Economics 112 (November 1997): 1203–50.
25. William Easterly, “Can Institutions Resolve Ethnic Conflict?” World Bank
Policy Research Working Paper No. 2482, February 2000. See www.eldis.org.
26. Brad Lian and John R. Oneal, “Cultural Diversity and Economic Development: A Cross-National Study of 98 Countries, 1960–85,” Economic Development
and Cultural Change 46 (October 1997): 61–77.
27. In partial correlations that control for log GDP at Purchasing Power Parity
in 1998, none of the four indicators of social diversity used in the present article
are correlated at a statistically significant level with annual average growth rate of
GDP per capita during the period 1990–99. Source of data on economic growth
rates: United Nations Development Programme, Human Development Report 2001
(New York: Oxford University Press, 2001), 178–81.
News and Notes
Two meetings were held in connection with articles published in the
April 2004 issue of the Journal of
Democracy. On March 8, the Forum
hosted a luncheon seminar entitled “Is
Anti-Americanism a Threat to Democracy?” featuring Ivan Krastev, chairman
of the board of Bulgaria’s Center for
Liberal Strategies. And on May 25, the
Forum hosted a speech by Francis
Fukuyama, professor of international
political economy at the Johns Hopkins
University, on “The Imperative of StateBuilding.” The event marked the
publication of his new book, StateBuilding: Governance and World
Order in the 21st Century.
In April and May, the International
Forum welcomed four new ReaganFascell Democracy Fellows:
Mohamed Al-Yahyai (Oman),
Oleksandr Fisun (Ukraine), Maria
Lisitsyna (Kyrgyz Republic), and
Fidaa Shehada (Palestinian Territories). A number of Reagan-Fascell
fellows gave presentations this spring:
On March 23, a roundtable discussion “From NAFTA to CAFTA:
Prospects for Strengthening Free
Trade and Democracy in Central
America” featured Francisco Villagrán
de León, former Guatemalan ambassador to the OAS and the UN.
A March 25 seminar entitled “The
New Face of Power in Azerbaijan: Assessing Ilham Aliyev’s First Hundred
Days in Office” featured Shahin
Abbasov, deputy editor-in-chief of
the Baku independent daily Echo.
On April 8, a luncheon seminar entitled “Armenia in Regional Context:
Prospects for Democracy and Integration” featured human rights activist
Anahit Bayandur, a former member
of the Armenian parliament.
191
An April 14 luncheon entitled
“Singapore: Myth or Model?” featured
Chee Soon Juan, secretary-general of
the Singapore Democratic Party and
director of the Open Singapore Centre.
On April 21, Vladimir Tismaneanu,
professor of government at the University of Maryland-College Park,
gave a presentation entitled “Democracy Romanian Style: Assessing Fifteen Years of Postcommunist
Transition.”
A May 18 luncheon event entitled “Is Mongolian Democracy in
Danger of Backsliding?” featured a
presentation by Mongolian democratic activist Enkhtuya Oidov.
On May 26, Lyudmila Georgieva,
founding chair of the Sofia-based
Foundation Common Cause, spoke
on “Policy Advocacy in Bulgaria.”
A June 1 luncheon presentation entitled “Pakistan: Democratization,
Authoritarianism, and the Consolidation of Military Rule,” featured
Pakistani political analyst and columnist Aqil Shah.
Correction
In the article “Does Diversity Hurt
Democracy?” by M. Steven Fish and
Robin S. Brooks in the January 2004
issue, there was an error in Table 1—
Regressions of Freedom House
Ratings on Hypothesized Determinants. The contents of the two rows
labeled Adj. R2 and N should be reversed, and every three columns in
the table should appear consolidated
in the N row. The corrected table is
available on the Journal website at
www.journalofdemocracy.org/Articles/FishandBrooksTable-15-1.pdf.